English Version

Features

1. Support lmdeploy for inference
2. Support training for Internvl2 Video
3. Support argument for LoRA target module in regex format
4. Support RTD website
5. Support argument of device_max_memory to config device_map memory usage.

New Models

1. Support CogVLM2-Video

中文

新功能

1. 支持lmdeploy框架的推理能力
2. 支持InternVL2模型的视频训练能力
3. 支持lora_target_regex参数，用来以正则表达式配置lora模块
4. 支持RTD文档网站
5. 支持device_max_memory参数来配置device_map的显存使用

新模型

1. CogVLM2-Video

What's Changed

• fix bug by @tastelikefeet in https://github.com/modelscope/swift/pull/1320
• fix template by @Jintao-Huang in https://github.com/modelscope/swift/pull/1321
• support cogvlm2-video by @Jintao-Huang in https://github.com/modelscope/swift/pull/1318
• Fix bugs by @tastelikefeet in https://github.com/modelscope/swift/pull/1325
• fix web-ui by @tastelikefeet in https://github.com/modelscope/swift/pull/1327
• compatible with trl 0.9.6 by @hjh0119 in https://github.com/modelscope/swift/pull/1326
• compat with vllm==0.5.1 by @Jintao-Huang in https://github.com/modelscope/swift/pull/1329
• Update qrcode by @tastelikefeet in https://github.com/modelscope/swift/pull/1332
• fix florence model by @hjh0119 in https://github.com/modelscope/swift/pull/1334
• Relaxing requirements for trl by @hjh0119 in https://github.com/modelscope/swift/pull/1342
• fix xcomposer2.5 device_map by @Jintao-Huang in https://github.com/modelscope/swift/pull/1343
• support generation_info by @Jintao-Huang in https://github.com/modelscope/swift/pull/1344
• fix requirements by @Jintao-Huang in https://github.com/modelscope/swift/pull/1347
• readthedocs by @tastelikefeet in https://github.com/modelscope/swift/pull/1345
• fix sequence parallel get labels by @tastelikefeet in https://github.com/modelscope/swift/pull/1352
• fix filelock by @tastelikefeet in https://github.com/modelscope/swift/pull/1354
• Add pt command by @tastelikefeet in https://github.com/modelscope/swift/pull/1356
• fix generation_info efficiency by @Jintao-Huang in https://github.com/modelscope/swift/pull/1359
• fix sh ddp_backend by @Jintao-Huang in https://github.com/modelscope/swift/pull/1360
• support LLM & lmdeploy by @Jintao-Huang in https://github.com/modelscope/swift/pull/1272
• fix a file path by @tastelikefeet in https://github.com/modelscope/swift/pull/1363
• Internvl2 support video by @tastelikefeet in https://github.com/modelscope/swift/pull/1366
• fix openai api by @Jintao-Huang in https://github.com/modelscope/swift/pull/1367
• fix internvl2-40b by @hjh0119 in https://github.com/modelscope/swift/pull/1369
• fix vlm deploy lora & agent by @Jintao-Huang in https://github.com/modelscope/swift/pull/1371
• Support lora regex by @tastelikefeet in https://github.com/modelscope/swift/pull/1375
• Fix docs by @Jintao-Huang in https://github.com/modelscope/swift/pull/1380
• Fix FSDP; Add training percentage to jsonl logging; Add a web-ui component by @tastelikefeet in https://github.com/modelscope/swift/pull/1381
• Support max memory args by @tastelikefeet in https://github.com/modelscope/swift/pull/1382
• fix max_memory by @tastelikefeet in https://github.com/modelscope/swift/pull/1383
• Fix gpu assert calculation by @tastelikefeet in https://github.com/modelscope/swift/pull/1384
• fix dataset_sample & deploy stop_words by @Jintao-Huang in https://github.com/modelscope/swift/pull/1385

Full Changelog: https://github.com/modelscope/swift/compare/v2.2.1...v2.2.2